[Suggestion] Supabase Edge Function serverless audio transcription (TypeScript)

[deepgram-robot]

What to build

A Supabase Edge Function that accepts audio file uploads (or URLs) and returns Deepgram transcriptions, with a companion web UI that lets users upload audio and see results — demonstrating serverless STT without managing infrastructure.

Why this matters

Supabase has 75,000+ GitHub stars and is one of the most popular backend-as-a-service platforms. Developers building on Supabase need a zero-infrastructure path to add transcription to their apps. Edge Functions run on Deno Deploy, providing global low-latency execution. This example shows the complete pattern: upload audio to Supabase Storage → trigger Edge Function → transcribe with Deepgram → store results in Supabase database.

Suggested scope

• Language: TypeScript (Deno runtime)
• Framework: Supabase Edge Functions, Supabase Storage, Supabase Database
• Deepgram APIs: Pre-recorded STT (Nova-3), Audio Intelligence (summarization)
• Pattern: Audio upload → Supabase Storage → Edge Function trigger → Deepgram STT → results stored in Supabase table
• Complexity: Medium — requires Supabase project setup, Edge Function deployment, and storage integration

Acceptance criteria

• Runnable with minimal setup (clone, add API keys, supabase start)
• README explains the pattern clearly
• Uses current SDK version
• Edge Function handles both file upload and URL transcription
• Includes Audio Intelligence summarization in the response
• Companion HTML/JS UI for testing uploads
• SQL migration for the transcription results table

---

Raised by the DX intelligence system.